A fast genetic algorithm for a critical protection problem in biomedical supply chain networks

In this paper, we present a new bilevel model for a biomedical supply chain network with capacity and budget constraint due to the protection and interdiction operations. The components considered in this model are biomedical devices, distribution centers (DCs), medical suppliers (MSs), and hospitals and patients as the demand points. On the other hand, two levels of decisions in the network planning is suggested: (1) the defender’s decision about protection operations of MSs and DCs, the assignment of clients to the DCs, and quantity of products shipped to DCs from MSs to minimize the demand-weighted traveling costs and transport costs and (2) the attacker’s decision about interdiction operations of MSs and DCs to maximize the capacity or service reduction and losses. Because of nondeterministic polynomial time (NP)-hardness of the problem under consideration, an efficient and fast approach based on a genetic algorithm and a fast branch and cut method (GA–FBC) was developed to solve the proposed model. Also, the efficiency via the comparison of results with the genetic algorithm based on CPLEX (GA-CPLEX) and decomposition method (DM) is investigated. In order to assess the performance of the presented GA–FBC, a set of 27 instances of the problem is used. Comprehensive analysis indicates that the proposed approach significantly solves the problem. In addition, the benefits and advantages of preference with running times and its accuracy is shown numerically. Simulation results clearly demonstrate that the defender’s objective effectively reduced and CPU time also within the large-sized instances of the problem in comparison with the GA-CPLEX and DM.     

A robust locating multi-optima approach for damage identification of plate-like structures

This paper proposes a robust optimization approach for multiple damage identification of plate-like structures. Different from traditional particle swarm optimizations (PSOs), a combined PSO and niche technique (NPSO) is proposed to solve multimodal optimization problems, with the full consideration of subswarm creation, merging and absorbing mechanism. As a hypersensitive parameter to damage, the curvature mode shape is adopted to construct the objective function. Case studies are conducted to investigate the effectiveness and robustness of the algorithm on multi-damage identification. Simulation results show that the proposed algorithm exhibits robust search performance on identifying damage locations accurately with good convergence behavior. It is hoped that this study can provide guidance on robust damage detection, especially when the structure is subject to multiple damages and external disturbances.     

Modeling of compressive strength of cemented sandy soil

This paper attempted to show the application of particle swarm optimization in the prediction of the compressive strength of cement sandy soil from the curing period, porosity of sample and percentage of cement. The results of the study show that the unconfined compressive strength of the cement stabilized sandy soil increases with an increasing cement content curing time period. Moreover the compressive strength decreases with an increasing porosity. The compressive strength improvement due to cement treatment has a larger increase in samples with less porosity. In addition, particle swarm optimization algorithm is and accurate technique in estimation of compressive strength of cement stabilized sandy soil. In order to compare of existing correlations, a total number of 100 unconfined compressive tests and 15 scanning electron microscope tests have been conducted on cemented Babolsar sand. It can be concluded that compared to existing correlations models, particle swarm optimization algorithm models give more reliable prediction about compressive strength of cement satblized sandy soil. Moreover, the sensitivity analysis of the polynomial model shows that cement content and porosity have significant impact on predicting unconfined compressive strength.     

大规模生物网络马尔可夫聚类的并行化算法

马尔可夫聚类算法（MCL）是在大规模生物网络中寻找模块的一个有效方法，能够挖掘网络结构和功能影响力较大的模块。算法涉及到大规模矩阵计算，因此复杂度可达立方阶次。针对复杂度高的问题，提出了基于消息传递接口（MPI）的并行化马尔可夫聚类算法以提高算法的计算性能。首先，生物网络转化成邻接矩阵；然后，根据算法的特性，按照矩阵的规模判断并重新生成新矩阵以处理非平方倍数矩阵的计算；其次，并行计算通过按块分配的方式能够有效地实现任意规模矩阵的运算；最后，循环并行计算直至收敛，得到网络聚类结果。通过模拟网络和真实生物网络数据集的实验结果表明，与全块集体式通信（FCC）并行方法相比，平均并行效率提升了10个百分点以上，因此可以将该优化算法应用在不同类型的大规模生物网络中。     

基于分解的多目标花朵授粉算法

在过去几十年里，许多多目标进化算法被广泛应用于解决多目标优化问题，其中一种比较流行的多目标进化算法是基于分解的多目标进化算法(MOEA/D)。花朵授粉算法是一种启发式优化算法，但迄今为止，花朵授粉算法在基于分解的多目标进化算法领域的研究还非常少。本文在基于分解的多目标进化算法的框架下，将花朵授粉算法拓展至多目标优化领域，提出一种基于分解的多目标花朵授粉算法(MOFPA/D)。此外，为了保证非支配解的多样性，本文提出一种基于网格的目标空间分割法，该方法从找到的Pareto最优解集中筛选出一定数量且分布均匀的Pareto最优解。实验结果表明，基于分解的多目标花朵授粉算法在收敛性与多样性方面均优于基于分解的多目标进化算法。     

Docker容器化下的遥感算法程序集成方法

目的 近年来，随着我国遥感技术的快速发展，遥感数据呈现出大数据的特点，遥感数据的时效性增强，针对新环境下遥感算法编程语言众多，程序运行和部署环境需求多样，程序的集成和部署困难的问题，提出了一种遥感算法程序快速封装与Docker容器化系统集成架构。方法 该系统架构主要包括：1）遥感算法程序的镜像自动化封装制作；2）镜像的分发管理，达到算法程序镜像的共享；3）遥感信息产品生产流程的容器化编排服务，将相关联的算法程序镜像串联，以满足特定遥感信息产品的生产；4）容器的调度运行，调用镜像，实现特定遥感产品的容器化运行。本文在上述容器化系统集成架构下，以Landsat5数据的NDVI、NDWI信息产品的生产作为容器化生产实例，并同物理机、KVM （kernel-based virtual machine）虚拟机在运行时间、内存占用量、部署效率等性能进行了对比。结果 Docker容器虚拟化环境下的产品生产和物理机环境下在运行时间和内存占用量上几无差别，优于KVM虚拟机。Docker容器虚拟化环境和KVM虚拟机环境下在部署上能够节省大量时间，相比于物理机环境能够提高部署效率。结论 容器化的系统集成方式能够有效解决遥感算法程序集成和部署困难的问题，有利于遥感算法程序的复用和流程的共享，提高系统集成效率，具备较强的遥感数据实时快速处理能力。     

双层过道布置问题的混合整数非线性规划模型及两阶段改进模拟退火算法

结合实际布局活动中，设施在多层空间布置的实际情况，提出了双层过道布置问题，并构建了该问题的混合整数非线性规划（MINLP）模型。针对问题特征，提出了一种改进模拟退火算法，通过采用两阶段改进策略，对退火过程及抽样过程进行改进。该算法以自适应搜索策略替代马氏链搜索长度，并引入记忆功能、回火操作以提高求解质量，通过设置双阈值来提高求解效率。应用所提算法对24个基准算例进行测试，并将其与原模拟退火算法和GUROBI精确方法作对比，验证了所提算法的有效性。     

采用信息增益率的混合入侵检测模型设计

针对现有混合入侵检测模型仅定性选取特征而导致检测精度较低的问题，同时为了充分结合误用检测模型和异常检测模型的优势，提出一种采用信息增益率的混合入侵检测模型.首先，利用信息增益率定量地选择特征子集，最大程度地保留样本信息；其次，采用余弦时变粒子群算法确定支持向量机参数构建误用检测模型，使其更好地平衡粒子在全局和局部的搜索能力，然后，选取灰狼算法确定单类支持向量机参数构建异常检测模型，以此来提高对最优参数的搜索效率和精细程度，综合提高混合入侵检测模型对攻击的检测效果；最后，通过两种数据集进行仿真实验，验证了所提混合入侵检测模型具有较好的检测性能.     

基于GPU加速的锥束CT重建算法研究

锥束计算机断层扫描（Cone-Beam Computed Tomography，CBCT）具有采集速度快和空间分辨率高等特点，被生物医学等领域广泛关注。然而通过CPU串行处理CBCT重建中海量投影数据非常耗时，难以满足实时性的需求。GPU的发展为CBCT重建的并行加速提供了条件。根据三角函数周期性的特点对FDK算法进行了改进，并利用GPU实现了12幅投影数据同时并行计算。实验结果表明，相比于传统基于CPU的重建算法，基于GPU的CBCT重建算法在保证图像质量的前提下，将重建速度提高了超过310倍。     

高斯混合模型结合加权似然的目标跟踪算法

鉴于高斯混合模型对背景变化快时无法精确检测出目标和目标跟踪的适应性差等瑕疵，提出了基于加权似然跟踪器来改进高斯混合模型实现运动目标跟踪算法。主要引入了自适应高斯混合模型来实时检测运动目标，然后空间加权似然来进行视频中的目标定位，引入加权似然期望值来改进高斯混合模型处理视频中的多尺度、多角度变化的目标跟踪不精准问题。通过VOT 2014 dataset对比实验结果表明提出的基于加权似然跟踪（Weighted Likelihood Tracking，WLT）和改进高斯混合模型（Improved Gaussian Mixture Model，IGMM）的目标跟踪算法较传统高斯混合模型跟踪算法在跟踪的精度有较大提高。在应对多尺度、多角度变化的目标跟踪表现出了较大的优势。